resource
Stranded Energy
Renewable generators produce on nature's schedule, not the grid's. When supply outruns demand, the energy gets curtailed, spilled, or sold at a loss. Bitcoin mining is the load that fits the shape: interruptible, location-agnostic, price-flexible. This is what that argument looks like in mechanics and data, not slogans.
The duck has a problem
Solar floods the California grid every spring afternoon. Net load — demand minus the renewables sitting on top of it — collapses around noon and ramps back up at sunset. Operators call the shape the duck curve. The belly is what we're here to talk about.
Representative spring day (April). Net load = grid demand minus solar & wind output. Negative values mean renewables exceed demand.
Toggle the mining-load overlay. It fills the belly. Whether that should happen — and what it changes for the project that built the panels — is the rest of the page.
Curtailment, by the month
Curtailment is the polite word for power that gets generated and then deliberately wasted. The grid operator tells the wind farm to feather its blades, or instructs the solar inverter to clip output, because there is nowhere for the electrons to go. CAISO has now curtailed roughly 7.8 TWh across the last 24 months — and the trend isn't flattening.
Two things compound here. The first is that solar build keeps outpacing transmission build, so spring midday and overnight wind increasingly land on a grid that can't absorb them. The second is that, in markets like ERCOT, the production tax credit makes it rational for wind operators to keep generating even when the locational marginal price goes negative — they are still better off after subsidies. So they pay the grid to take their power.
ERCOT spent more than 1,500 hours below $0/MWh at one or more nodes in 2024. That is paying the grid, not being paid by it.
The storage wall
The intuitive answer to curtailment is storage. Add batteries; charge when there's surplus; discharge when there isn't. The intuitive answer is half right. Batteries handle the daily-cycle mismatch reasonably well. The longer the duration, the worse the economics.
Lithium-ion solves the four-hour evening ramp. It does not solve the spring-glut-to-winter-deficit problem. Different problem, different physics.
Battery cost scales with energy capacity, not power. Doubling the duration roughly doubles the price. Utilization drops at the same time, so cost-per-cycle climbs.
Mining doesn't compete with batteries. It absorbs the energy that batteries leave on the table — the seasonal surplus, the transmission-stranded MWh, the negative-price hours.
Three load properties grids actually want
Most loads on the grid behave like furniture: they show up, they consume, they don't move and they don't bend. Mining is different in three specific ways, and the three are what make it the right buyer for stranded power.
Interruptible
Sheds load in seconds.
A mining facility can drop tens of megawatts off the grid faster than any natural-gas peaker can spin up. ERCOT pays demand-response participants for exactly this responsiveness; mining is a near-perfect demand-response asset that already exists at scale.
Location-agnostic
Needs power and a network connection. That's it.
Aluminum smelters need ports, data centers need fiber backbones, factories need workers. Mining only needs power and a thin internet pipe. You put it next to the generator instead of next to a city — sidestepping the transmission queue that strands most renewable projects.
Price-flexible
Voluntarily off below break-even, on above.
Below the hashprice that covers its electricity cost, a rational miner turns off. Above it, a miner buys every available watt. That makes mining the kind of price-elastic load grid operators have wanted for decades and never had — a buyer that bids by formula, not by habit.
Buyer for the energy nobody else wants
For an existing renewable project, the question isn't whether mining is good or bad in the abstract. It's whether co-locating a mining load improves the IRR enough to justify the integration work. The simulator below is small but defensible: it takes the four numbers that actually move the answer and shows what they do.
Three projects in the wild
Project archetypes, not branded case studies. The third is a fossil example included because the same mechanism applies — flagged as such, not laundered.
ERCOT wind, West Texas
West Texas built wind faster than transmission could carry it east. Overnight wind regularly drives nodal prices below zero — generators pay the grid to take their power because the production tax credit makes negative pricing rational.
Co-located mining loads buy power locally during the negative-price window, then shed within seconds during ERCOT's four-coincident-peak hours under demand-response programs.
Mining provides firm bid for energy that would otherwise be curtailed and is paid by ERCOT for being interruptible during peak. The wind project's effective realized price improves; the grid gets a flexible load.
Run-of-river hydro with seasonal overflow
Spring snowmelt and high-water months produce more generation than the local grid can absorb or transmit. Operators spill water past the turbines — energy literally pours over the dam.
Mining containers sited at the powerhouse absorb the seasonal surplus. When river flow drops or local load rises, the miners ramp down within minutes.
Spilled-water months become revenue months. The dam operator captures value from generation that previously had nowhere to go, without affecting the firm-power contracts the dam already serves.
Associated gas at remote oilfields
Oilfields in basins without sufficient gas takeaway capacity flare the associated methane that comes up with the oil. Flaring is a regulatory and emissions liability; venting is worse.
On-site generators burn the otherwise-flared gas to power miners at the wellhead. Combusting methane in a generator is a meaningful improvement over flaring it directly: methane is ~80x more potent than CO₂ over twenty years, and combustion is more complete in an enclosed engine than in an open flare.
The oil producer earns revenue on a stream that was a liability. Field-level methane intensity drops. The pattern is associated with one company (Crusoe) more than any other; flagged here as the one branded reference because the stranded-gas mechanism is inseparable from them.
Stranded fossil energy, not renewable. Included for the same mechanism: an interruptible, location-agnostic buyer monetizes energy that would otherwise be wasted.
The strongest counter is that mining's emissions footprint is a function of its grid mix, and a miner that buys curtailed renewables today can buy fossil baseload tomorrow. True. The response is that the location-agnostic property cuts both ways: mining sites where power is cheap, and curtailed renewables consistently produce the cheapest hours on the grid. The economic gradient pulls toward stranded clean power, not away from it. That's a tendency, not a guarantee — worth saying out loud.
The bridge for projects that don't pencil yet
Section 5 was about mature projects with curtailment to monetize. This is the smaller, harder case: a renewable project sited where the resource is excellent and the transmission is years away. Without an off-taker, the project doesn't finance. With mining as the day-one buyer, it can.
Generation comes online. Transmission upgrade is years out. No PPA buyers can take the power. Mining absorbs 100%.
Mining's role shrinks as the project matures into the grid. That's the point. It bridges the revenue gap between the generator coming online and the transmission queue clearing — five years that otherwise kill the deal at financing. Then it gracefully exits to the same last-resort role as everywhere else.
The system already throws away enough renewable energy to power tens of millions of homes. Some of it gets monetized by an interruptible, location-agnostic, price-flexible buyer that already exists at scale. The rest doesn't. That's the whole argument.